Applied Energy 88 (2011) 1366–1376

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Wet and dry cooling systems optimization applied to a modern

waste-to-energy cogeneration heat and power plant
G. Barigozzi, A. Perdichizzi, S. Ravelli ⇑
Department of Industrial Engineering, Bergamo University, Italy

a r t i c l e i n f o a b s t r a c t

Article history: In Brescia, Italy, heat is delivered to 70% of 200.000 city inhabitants by means of a district heating system,
Received 9 January 2010 mainly supplied by a waste to energy plant, utilizing the non recyclable fraction of municipal and indus-
Received in revised form 1 August 2010 trial solid waste (800,000 tons/year, otherwise landfilled), thus saving annually over 150,000 tons of oil
Accepted 27 September 2010
equivalent and over 400,000 tons of CO2 emissions.
Available online 29 October 2010
This study shows how the performance of the waste-to-energy cogeneration plant can be improved by
optimising the condensation system, with particular focus on the combination of wet and dry cooling sys-
Keywords:
tems.
Wet and dry condenser
Cooling tower
The analysis has been carried out using two subsequent steps: in the first one a schematic model of the
Air condenser steam cycle was accomplished in order to acquire a knowledge base about the variables that would be
Off-design optimization most influential on the performance. In the second step the electric power output for different operating
conditions was predicted and optimized in a homemade program. In more details, a thermodynamic
analysis of the steam cycle, according to the design operating condition, was performed by means of a
commercial code (ThermoflexÓ) dedicated to power plant modelling. Then the off-design behaviour
was investigated by varying not only the ambient conditions but also several parameters connected to
the heat rejection rate, like the heat required from district heating and the auxiliaries load. Each of these
parameters has been addressed and considered in determining the overall performance of the thermal
cycle. After that, a complete prediction of the cycle behaviour was performed by simultaneously varying
different operating conditions. Finally, a MatlabÓ computer code was developed in order to optimize the
net electric power as a function of the way in which the condensation is operated. The result is an opti-
mum set of variables allowing the wet and dry cooling system to be regulated in such a way that the max-
imum power is achieved. The best strategy consists in using the maximum amount of heat rejection in
the wet cooling system to reduce the operational cost of the dry one.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction the latter, the cooling water taken from the condenser passes
through a wet mechanical draft tower (CT) and returns to the con-
The cooling system in a power plant rejects heat at approxi- denser. The obvious advantage of a closed-loop system is the
mately twice the rate at which electric power is generated [1]. It reduction in the water demand at the expense of larger operational
is important to underline that, since energy flows in power plants costs. Nevertheless environmental regulations and the increasing
are typically high, small improvements to the cooling system can scarcity of natural water supplies make both the AC and the CT
lead to large fuel savings and consequently efficiency enhancement convenient [3].
[2]. The effectiveness of a cooling system can be quantified through The way in which the heat rejection takes place in an AC does
the condensing steam pressure: the lower the pressure the greater not need to be deeply investigated: the critical element of the plant
the effects. is the CT. Although the basic parts of the tower can easily be de-
The present study was inspired by the operation of a waste-to- scribed and understood, heat and mass transfer processes are very
energy cogeneration plant placed in Brescia, Northern Italy (Fig. 1). complex. The temperature of the circulating water is reduced by
Attention was drawn to the condensing units because of the com- bringing it into direct contact with air: the cooling is attained
bined wet and dry cooling system (Fig. 2): it is composed of an air partly by the evaporation of a fraction of the water flow rate and
cooled condenser (AC) in parallel with a water cooled condenser. In partly by the transfer of sensible heat [4]. Each water particle is
surrounded by a film of saturated air which is considered to be
⇑ Corresponding author. Tel.: +39 035 2052346; fax: +39 035 2052077. at the same temperature as the water; the air is heated and be-
E-mail address: [email protected] (S. Ravelli). comes saturated with moisture as it passes through the tower. In

0306-2619/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2010.09.023
 G. Barigozzi et al. / Applied Energy 88 (2011) 1366–1376 1367

Nomenclature

AC air condenser LP low pressure

AUX auxiliaries m flow rate
CT cooling tower n rotational speed
CWP cooling water pump P net power
d design RH relative humidity
DH district heating Tamb ambient temperature
HP high pressure W&D wet and dry layout

the heat transfer process more then two-thirds of the heat is trans- of differential equations [5–8], with the help of CFD modelling
ferred by evaporation with the rest being transferred by convec- [9,10], or by the NTU methods [11,12].
tion. In many works a great effort has been made to model the All cited papers think about the CT as a standalone component.
physical situation within a CT: since films and droplets of water According to different viewpoints it could be considered as a device
are in constantly changing configuration there is no theoretical which can minimize the environmental impact in the process
investigation capable of simulating such a complex phenomenon industry thanks to the recirculation of the cooling water [13,14]
without simplifying assumptions. It could be described by a set or increase the efficiency of a power plant [15].
The present study complies with the last approach and wants to
show the effect of a wet and dry condensation system on the per-
formance of a cogenerative plant. At author’s knowledge, no works
concerning this field can be documented. This paper offers an ori-
ginal contribution not only for considering a split system for air
condenser and wet cooling tower but also for defining an optimiza-
tion procedure whose aim is to maximize the net electric power for
any operating condition. Consequently the power consumption of
all auxiliaries was accurately estimated. It is necessary to remem-
ber that a mechanical draft CT uses fans to provide the required
volume of airflow: their cost is an additional expense to be consid-
ered. With respect to a natural draft CT, where the circulation is en-
sured by the density difference existing between the heated, moist
air and the fresh air, the mechanical draft CT can be built with rel-
atively less expensive materials but requires higher operation and
maintenance costs. The major problems are associated with fog-
ging and recirculation. However the risk of recirculation is reduced
if the fan is located on the top of the tower, because of the high dis-
charge velocity. Furthermore, pumps with variable speed are re-
quired in order to guarantee the most favourable mass flow rate
of cooling water. Another issue dealing with the operation of a
Fig. 1. View of the waste-to-energy cogeneration plant placed in Brescia, Northern
CT is the plume formation: when the atmosphere is too cold and
Italy. humid to absorb the moisture in the exhaust air from the CT, it fol-

Fig. 2. Schematic of a split system for air condenser and wet cooling tower [16].
1368 G. Barigozzi et al. / Applied Energy 88 (2011) 1366–1376

lows that the excess moisture condenses and the water droplets Table 1
become visible. Nevertheless the plume is not a pollutant, many Reference cycle design parameters.

ways to avoid it are well documented in technical papers [16]. In Ambient temperature (°C) 15
spite of that, plume abatement is not widely applied since it is a Ambient pressure (bar) 1.01325
very expensive feature. Ambient R.H. (%) 60
HP turbine inlet pressure (bar) 72
HP turbine inlet temperature (°C) 450
HP turbine inlet steam flow rate (t/h) 345
2. Power plant and design operating conditions
DH heat load (MW) 22.8
AC fan load (%) 100
The present investigation refers to the thermal cycle of a waste- CT fan load (%) 100
to-energy 80 MWe cogeneration plant. The flue gas from the com- Water cooled condenser pumps load (%) 80
bustion of the municipal solid waste, downstream of materials Gross power output (MW) 81.5

recovery, enters the heat recovery steam generator to produce

steam. This steam then passes through a steam turbine, as shown
in Fig. 3: the turbine is split up into a high pressure (HP) section of 345 t/h entering the HP turbine at 72 bar and 450 °C. In the de-
and a low pressure (LP) one, by allowing the steam extraction for sign condition the steam fraction going to the DH is about 9.56% of
district heating (DH). The remaining steam flow rate goes to con- the turbine inlet steam flow rate, supplying 22.8 MWt. Steam frac-
densation passing into an AC or into a water condenser. The heat tion going to condensation is about 76% of turbine inlet steam mass
transferred between the steam side and the cooling water is re- flow rate.
jected in a mechanical draft CT. Between high and low pressure turbines, a crossover valve (1) is
Two pumps in parallel connection ensure the adequate circula- used to control steam extraction and HP turbine downstream pres-
tion of the cooling water in a closed loop. All pumps and fans oper- sure. A second valve (2) placed before the water cooled condenser
ate under inverter control so that their rotational speed can be is closed when ambient temperature is under 10–12 °C (i.e. pro-
continuously varied. vided that the AC is able to fully condense the steam), so all the
Finally the two condensate flows mix together and go to the steam flow rate goes to the AC. This plant layout has been defined
deareator. The elements marked with letters a–e symbolize the as ‘‘dry” (Dry), with reference to the cooling system, in opposition
connection between the cycle components through pipes. Each to the ‘‘wet and dry” configuration (W&D) described before.
pipe is characterised by a specific pressure drop, mainly depending
on its geometric features.
The scheme presented in Fig. 3 included some simplifications in 3. Simulation method
comparison with the real plant. The heat recovery steam generator
and the feed water heater were not considered since the analysis Heat rejection to the ambient, net power output and cycle effi-
was focused on the condensation system, with a steam production ciency can vary because of atmospheric conditions (temperature
at a constant mass flow rate, pressure and temperature. Moreover, and RH), implying a modification in both AC and CT behaviour,
a simplified modelling of the steam extractions from the HP tur- but also because of the thermal load required by DH. The final ef-
bine was performed: a single steam extraction feeding both DH, fect is a change in the steam flow rate going through the con-
feed water heaters and combustion air pre-heaters was used in- denser. Moreover, the LP turbine back pressure (and the HP one,
stead of modelling three different extractions. Anyway, the model even if in a limited way) can be changed by a different control
assured the same steam flow rate entering the LP turbine as in real strategy of the condensation system. When a W&D configuration
operating conditions. is used, this control strategies may be complicated, involving sev-
Table 1 reports the main cycle parameters at design conditions. eral parameters. In fact, heat rejection can be varied changing both
The analysis has been carried out for a constant steam production AC and CT fans rotational speed as well as cooling water pumps

Fig. 3. Thermal cycle layout with the Wet and dry cooling system.
 G. Barigozzi et al. / Applied Energy 88 (2011) 1366–1376 1369

rotational speed through inverters. An optimization strategy was Table 2

thus developed in order to find out the best control strategy, i.e. LP turbine design data.

the combination of fans and pumps rotational speeds leading to Inlet pressure (bar) 1.118
the maximum net power output, whatever the operating condi- Inlet steam flow rate (t/h) 281.8
tions may be. Exit enthalpy before exhaust loss (kJ/kg) 2313.5
Dry exhaust loss (kJ/kg) 19.34
This optimization process was carried out by an in house devel- Exhaust volume flow (m3/s) 602
oped MatlabÓ code, taking the net power output P as the objective Annulus area (m2) 3.06
function. The user is first required to choose the plant configura-
tion (Dry or W&D), the ambient conditions (temperature T and rel-
ative humidity RH), and the heat demand from DH. The program
first makes use of these input data to correct the net power output
from the design (Pd) to the off design point (P) by means of a few
coefficients (k). The net power P is then further corrected using
polynomial functions, each one taking into account for the influ-
ence of the following parameters: the AC fan speed nAC, the CT
fan speed nCT and the cooling water pump speed nCWP.
The implemented procedure attempts to find a constrained
minimum of a scalar function (P) of several variables (nAC, nCT
and nCWP), each one depending on ambient and DH conditions,
starting at an initial estimate (P0). The minimum is thus subjected
to the linear inequalities Ax 6 b. The resulting problem to be solved
consisted in

min funðPÞ ¼ kT amp kRH kDH Pd P3i¼1 ðAi n3i þ Bi n2i þ C i ni þ Di Þ ð1Þ

ni being the three already mentioned fluid machinery rotational

speeds, ki the correction factors for ambient conditions and DH
and Ai to Di the polynomial interpolation coefficients describing
the net power output P dependency on each influence parameter.
The code finally gives the AC fan speed, the CT fan speed and the Fig. 4. LP turbine leaving loss as a function of the exhaust volume flow.
cooling water pumps speed which optimize the net power
production.
In order to obtain a realistic simulation many details were included
Each considered variable is subjected to the constraints listed in
in the model. One of the most critical components to be sized was
Table 9. In particular, the ambient temperature can vary between
the LP turbine, whose design data are listed in Table 2. In fact, to
10 °C and 35 °C in the W&D configuration, a range typically
correctly predict its off-design behaviour, the exhaust losses were
encountered in Northern Italy in the warmest seasons (i.e. from
evaluated as a function of the exhaust volume flow (Fig. 4). The de-
April to October), and from 5 °C up to 15 °C in the Dry configura-
sign point is located close to the loss minimum condition, where it
tion, operating in the October–April time period. An overlap of 5 °C
is supposed to be. The inlet pressure control to LP turbine, defined
was maintained to check for the effective convenience of moving
as sliding, lets the pressure vary with the steam flow rate. On the
from W&D to Dry configurations. A 10–100% fans operational
contrary, pressure at the inlet of the HP turbine is fixed and does
range in the W&D case was considered, reducing to 40–100% for
not depend on the steam flow rate. Tables 3–5 summarized the de-
the Dry case. This was chosen to avoid freezing and to comply with
sign parameters for the most relevant cycle components (AC, water
the plant’s everyday routine operation. Also the range of the heat
cooled condenser and CT, respectively) and the corresponding siz-
demand from DH is limited, depending on the considered conden-
ing procedure results. In the water cooled condenser a condensate
sation asset: in the W&D layout the heat load varies between
subcooling of about 5 °C was introduced in order to match the
20 MW and 100 MW, while in the Dry layout it varies from
experimental data. This can be related both to a not proper design
60 MW up to 160 MW, resulting in a very large variation of mass
of this component and to the way in which it is connected to the LP
flow rate going to condensation.
turbine exhaust diffuser. In fact, a long channel connects the LP
In order to resolve the model for any possible operating condi-
turbine exit with the water cooled condenser, generating high
tion occurring during the year, thus defining both ki and polyno-
pressure losses, and probably a non uniform steam distribution
mial coefficients values, a detailed database containing the power
over the condenser heat transfer surfaces.
plant off-design thermodynamic performances was generated; an
Once all components were sized according to design values and
interpolation method was then used to rapidly calculate the poly-
their stand alone off-design performance verified against manufac-
nomial coefficients.
turers data, they were connected as reported in Fig. 3 to build up
the whole plant model, and operated in off-design conditions. Pres-
4. Power plant performance database sure drops in connection pipes were estimated from DCS data and

The power plant simulation software ThermoflexÓ, tuned with

performance data provided by the manufacturers, has been used Table 3
to build up the plant model necessary to generate the data base cy- AC design data.
cle performance, according to different operating conditions. Condenser pressure (bar) 0.095
It should be noted that each cycle component needed a specific Condensate subcooling (°C) 0
design according to features which did not necessarily match each Air temperature rise (°C) 16.8
other. A design procedure was carried out in order to define mass, Air flow rate (t/h) 21,215
Fan efficiency (%) 70
heat balances and sizing of each cycle component, followed by a
Fan power consumption (kW) 1160
comparison between simulation results and manufacturer data.
1370 G. Barigozzi et al. / Applied Energy 88 (2011) 1366–1376

Table 4 developed in order to understand the way each relevant parameter

Water cooled condenser design data. influences the net power output. The second goal was the assess-
Condenser pressure (bar) 0.13 ment of the superposition effect applicability to the present analy-
Condensate subcooling (°C) 4.8 sis. Finally, the optimization of net power output as a function of
Cooling water temperature rise (°C) 10.1 condensing system operational parameters was carried out for
Cooling water mass flow rate (t/h) 7370
Pump efficiency (%) 85
the two investigated condensing system configurations (Dry and
W&D), whatever ambient conditions and DH requirements.

Table 5 5.1. Model validation

CT design data.

Cooling tower approach to wet bulb temperature (°C) 6.2 The model was first run at the plant design condition, consid-
Air wet bulb temperature rise in wet section (°C) 12.2 ered as a reference, and also under different ambient conditions
Number of cells 3
and DH requirements, in order to evaluate its prediction capability.
Cooling water mass flow rate (t/h) 12,000
Air mass flow rate (t/h) 7695 Table 7 compares computed and measured data for three different
Fan power consumption (kW) 560.8 operating conditions of the W&D configuration. All the reported
cases refer to a full condensing asset, corresponding to the mini-
mum steam extraction for DH.
plant lay-out schemes (Table 6). The combination of wet and dry Table 8 compares the model results against the cycle perfor-
cooling systems implies that a portion of the exhaust steam may mance data provided by the turbine manufacturer, for the Dry con-
condense by the water condenser and another portion may con- figuration. The reported cases show three values of the steam flow
dense by air in the AC. A balancing splitter was chosen to govern rate entering the HP turbine and different DH requirements. In par-
the exhaust steam allocation so as to balance the turbine exit pres- ticular, case 1 refers to a steam extraction for DH of about 60% of
sure: the basic idea is that both the condensers, except of the pip- the HP turbine inlet flow, corresponding to an heat demand of
ing connections pressure drops, work at the same pressure level. about 55 MWt while, in case 2 and case 3, the steam extraction
for DH is relatively low (about 20% of HP turbine inlet flow), corre-
sponding to an heat demand of about 20 MWt. The model predic-
5. Results and discussion tion capability was verified by comparing computed and measured
LP turbine exhaust steam temperature values as well as AC, cooling
As a first step of this investigation, the power cycle model was water condenser and CT inlet and outlet temperature values. The
validated against operational data. Then a parametric analysis was net power production cannot be directly compared, due to the fact
that the HP turbine steam extractions were not simulated in a de-
Table 6 tailed way, leading to an overestimation of the computed HP tur-
Pressure drops in pipes (with reference to Fig. 2). bine power. Few other cases were considered to validate the
Pipes Pressure drop (dP/P) W&D model but they were not included in the paper. For all the
cases, a good agreement was found between model results and
(a) 0.08
(b) 0.07
measurements: the maximum difference is lower than ±8%, as
(c) 0.001 shown in Tables 7 and 8.
(d) + (e) 0.5 + 0.5 In the W&D case, the simulation provides also the steam shar-
ing between the two condensers: approximately 54% of the steam

Table 7
Comparison between model results (M.) and experimental data (Exp.) – W&D.

Variables Case 1 Case 2 Case 3

M. Exp. M. Exp. M. Exp.
Ambient temperature (°C) 22.1 16.4 14.5
Turbine inlet flow rate (t/h) 345 370 318
CT flow rate (t/h) 7437.5 7370 7350
Exhaust steam flow rate (t/h) 270 250 292 267 286 275
Exhaust steam temperature (°C) 43.8 44.7 41.2 43.1 39.4 40.5
Condensate temperature from AC (°C) 42.3 44.3 39.7 42.2 37.9 39.3
Cooling water temperature to the CT (°C) 30.8 32.6 28.1 27.7 25.6 27.1
Cooling water temperature from CT (°C) 22.3 25 19.6 20 18.8 20.5
Condensate temperature from water condenser (°C) 36.3 38.3 33.6 34.5 31.1 33
Steam temperature to water condenser (°C) 41.1 42.3 38.4 40.1 36.2 37.8
Steam flow rate to AC (t/h) 140 – 159 – 159 –

Table 8
Comparison between model results and cycle data provided by the turbine manufacturer – dry.

Case 1 Case 2 Case 3

Model Ref. Model Ref. Model Ref.
Turbine inlet flow rate (t/h) 212 245.5 370
Turbine inlet pressure (bar) 60 60 76.3
Steam flow rate to AC (t/h) 87.2 200.3 281.7
Exhaust steam temperature (°C) 35.5 35 46.7 47.5 56.5 57.1
 G. Barigozzi et al. / Applied Energy 88 (2011) 1366–1376 1371

flow rate goes to the air cooled condenser in the tested operating
conditions (i.e. with minimum steam extraction for DH). Unfortu-
nately, this information is not made available by DCS.

5.2. Parametric analysis

Afterwards a parametric analysis was accomplished, starting

from the design point (see Table 1), to assess the thermal cycle per-
formance over a complete range of operating conditions. Taking
into consideration both the plant layouts (W&D and Dry), several
test cases were executed by varying the parameters listed in Table
9. In particular, the ambient temperature Tamb was varied with 5 °C
steps in a range depending on the plant layout. Relative humidity
RH variation was only considered in the hottest climate, i.e. in
the W&D configuration. The AC and CT fan loads and pumps loads
were also included, ruling the heat transfer in both condensers and
in the CT. Finally, also the effect of the heat demand from DH was
evaluated. All ranges reported in Table 9 are large enough to cover
the whole plant operational field.
The study initially focused on the effects of each input variable
on some selected outputs (net power output P, auxiliaries power
consumption PAUX, LP turbine exhaust pressure pout and steam
fraction to AC, mAC/mLP) in order to verify the behaviour of each
component and if a superposition rule can apply at least to some
of the considered variables. Moreover, the numerous simulations
constitute a fundamental database to derive the mathematical
relation between the net power (object function of the optimiza-
tion process) and a combination of inputs.

5.2.1. W&D configuration

Figs. 5–9 show the influence of the aforementioned parameters
on some selected outputs from the W&D model. In each picture the
net power output is normalized using the design value, while AC
and CT fans load as well as cooling water pumps load are computed
as the ratio between the actual and the maximum fluid machinery
rotational speed.
Fig. 5a shows the normalized net power output variation versus
Tamb and RH. As expected, P/Pd decreases with rising RH. The effect
becomes more and more appreciable with increasing temperature.
This is consistent with the LP turbine exhaust steam pressure
behaviour shown in Fig. 5b: the condensing pressure increases Fig. 5. RH and Tamb influence on (a) normalized net power output and (b) exhaust
with ambient temperature making turbine power decrease, while steam pressure – W&D.
the auxiliary power consumption (not shown) remains practically
unchanged. In the worst condition, the combined effect of high noting that fan regulation according to the maximum load is not
RH = 100% and Tamb = 35 °C causes a large increase in turbine ex- the right way to get the highest power production: the power gain
haust pressure (up to 0.17 bar) and a significant reduction of about due to the reduction in condensing pressure is not enough to com-
2% in net power output with respect to design condition. pensate the increase in fan power consumption. So fan load rang-
The influence of the AC fan load on the cycle performance is ing from 30% to 80% was found to give the best performance. It is
shown in Fig. 6. An optimum fan load maximizing the net power obvious that the steam flow rate entering the AC increases with
output can be easily identified (Fig. 6a), whose value progressively raise in fan load and consequently in fan speed (Fig. 6d).
increases with Tamb. This optimum condition is a compromise be- A very similar plant behaviour was also found varying the CT
tween the auxiliaries power consumption increase (Fig. 6b), almost fan load (Fig. 7), thus controlling the heat rejection in the water
independent from ambient temperature, and the exhaust pressure cooled condenser. A CT fan load increase at constant ambient tem-
decrease (Fig. 6c), the latter strongly influenced by Tamb. It is worth perature initially gives rise to an exhaust pressure decrease
(Fig. 7c) that, coupled to a small PAUX increase (Fig. 7b), results in
a relevant P/Pd augmentation (Fig. 7a). Note that the CT fan power
Table 9 input (PCT,d = 561 kW) is less than half the AC fans one
Parametric analysis ranges. (PAC,d = 1160 kW). The progressive increase of PAUX counteracts
Input variables Wet and dry Dry the beneficial exhaust pressure decrease, that also stabilizes due
Ambient temperature (°C) 10–35 5–10
to increased leaving losses, giving the net power output a decreas-
Relative humidity (%) 20–100 20–100 ing trend. Also the full load operation of CT fan is not a good ap-
Air condenser fan load (%) 10–100 40–100 proach to achieve the maximum power. Even at the highest
Cooling tower fan load (%) 10–100 – investigated ambient temperature, the fan load should not exceed
Cooling water pumps load (%) 20–100 –
70%. Of course, the AC steam fraction decreases increasing the CT
DH heat load (MW) 20–100 60–160
fan load (Fig. 7d).
1372 G. Barigozzi et al. / Applied Energy 88 (2011) 1366–1376

Fig. 6. AC fan load and Tamb influence on (a) normalized net power output, (b) auxiliary power consumption, (c) exhaust steam pressure and (d) AC steam fraction – W&D.

The amount of heat exchanged in the water condenser can be creases. The steam distribution at constant temperature and var-
regulated also by varying the cooling water flow rate through a iable DH heat requirement instead depends on the ambient
change in pumps rotational speed (Fig. 8). The plant behaviour temperature level: at the lowest temperature of 10 °C the AC
for variable pump load is not different than before, as the effect steam fraction increases with rising DH heat load. The opposite
on exhaust pressure is almost unchanged. Fig. 8a shows that the occurs at the highest temperature of 35 °C. This behaviour is con-
choice of cooling water flow rate exceeding the design point (cor- sistent with the water cooled condenser better performances with
responding to an 80% pump load) produces an increase in the net rising air temperature.
power only for high temperature (35 °C). At low ambient temper-
ature the circulation pumps should be run at partial speed to re- 5.2.2. Dry configuration
duce the power consumption: a pump load lower than 33% is In the coldest period of the year (5–15 °C) the AC receives the
enough for a temperature range between 10 °C and 20 °C. whole exhaust steam flow rate. Thus the investigation has been re-
The influence of the DH heat demand on cycle performance in stricted to ambient temperature, to the AC fan load and to the heat
shown in Fig. 9. P/Pd linearly reduces with an increase in the heat demand from DH.
demand from DH up to 85 MW. This is mainly due to the reduc- Fig. 11 reports some of the parametric analysis results for the
tion in the power produced by the LP turbine, as a result of a Dry configuration. The effects of air temperature on net power out-
decreasing steam flow going into it, even if the LP turbine exhaust put are not reported as they are very similar to those given for the
pressure progressively decreases (Fig. 9b). Increasing the DH heat W&D layout, even though the influence is less marked, as a result
load further, P quickly decreases. This is due to the crossover of more favourable ambient conditions. On the contrary, the effect
valve closing strategy, resulting in a constant HP turbine exhaust of the AC fan load on net power output greatly differs from the
pressure, as reported in Fig. 10. Concerning the way in which the W&D behaviour: in fact, if Tamb varies between 5 °C and 15 °C,
exhaust steam is shared into the two condensers, Fig. 9c shows, the AC fan should be operated according to the maximum load to
as expected, that an increasing steam mass flow rate goes get the highest power production. The regulation of fans speed,
through the water cooling system as the ambient temperature in- so as to reduce the auxiliary power consumption, is convenient
 G. Barigozzi et al. / Applied Energy 88 (2011) 1366–1376 1373

Fig. 7. CT fan load and Tamb influence on (a) normalized net power output, (b) auxiliary power consumption, (c) exhaust steam pressure and (d) AC steam fraction – W&D.

in hard weather conditions. As far as the DH is concerned, no sig- of the AC fan load and the DH heat load or RH on the optimum net
nificant matters deserve notation; in fact the Dry layout reacts in power output.
the same way as the W&D layout to variation in the heat demand From the optimized AC and CT fans behaviour (Fig. 12) it can be
from DH. deduced that, if Tamb < 15 °C, the condensation heat is more effi-
ciently dissipated in the AC: in fact, in this condition, the optimized
AC fan load (60–70%) is higher than the CT fan load (45.7–47.5%).
5.3. Optimization of the cycle performance
On the contrary, if Tamb is hotter, the water cooling system ensures
the greatest results in terms of power production: the optimized
The performed parametric analysis demonstrated that the
CT fan load (86.8–100%) is now higher than the AC fan load (40–
superposition rules do not apply either to the W&D or to Dry con-
80%).
figuration. Nevertheless, this analysis provided a detailed data base
The optimization of the W&D layout led to the following major
for plant performance evaluation at different operating conditions.
suggestions:
This data base was then included in an optimization procedure
whose final target was the selection of a set of fans and pumps
– The pumps rotational speed should gradually increase with
loads to operate the condensing system in such a way that the
Tamb, passing from about 33% at 10 °C to 100% at 35 °C.
maximum power is achieved at different ambient conditions and
– Similarly, the CT fan rotational speed should be raised from 45%
DH heat demands.
to 87% while Tamb increasing up to 20 °C and maintained at the
highest level (100%) if Tamb exceeds 25 °C.
5.3.1. W&D configuration – Concerning the AC fan rotational speed, the range between 40%
Figs. 12 and 13 report the results of the condensing system opti- and 80% is confirmed to provide the best performance.
mization procedure for the W&D configuration. In particular,
Fig. 12 shows the optimum load of CT fan, AC fan and pumps by In the warmest conditions the plant should be regulated accord-
varying air temperature, while Fig. 13 shows the combining effects ing to the following logic: heat rejection has to be performed as
1374 G. Barigozzi et al. / Applied Energy 88 (2011) 1366–1376

Fig. 8. Cooling water condenser pumps load and Tamb influence on (a) normalized
net power output, (b) exhaust steam pressure – W&D.

much as possible by the cooling water; the rest of the condensation

heat has to be discarded in the AC. This complies with the auxiliary
power consumption: at full load, the AC fan uses more than twice
the CT fan electricity. So, to sum up, an increase in temperature
shifts condensation from the dry to the wet cooling system. The
shift point is between 15 °C and 20 °C.
The heat demand from DH and the ambient RH do not affect the
optimal point but the objective function value at the optimal point.
As shown in Fig. 13, the increase in both the DH heat load and the
RH causes the optimized net power scaling down, while maintain-
ing the same optimum condition, in terms of CT, AC, and pumps
load. This is valid whatever the ambient temperature may be.

5.3.2. Dry configuration

The optimization of the plant working according to the dry lay-
out required a smaller effort since the objective function depends
on one single variable. The AC fan load which maximizes the net Fig. 9. DH heat load and Tamb influence on (a) normalized net power output, (b)
exhaust steam pressure and (c) AC steam fraction – W&D.
power, by varying ambient temperature, is shown in Fig. 14. Unlike
the previous case, steam extraction to DH influences the optimal
point at intermediate temperatures (0–5 °C). In this range, the AC feeds the DH, otherwise it would be set to 80%. At warmer temper-
fan load can be reduced to 60% if a great part of the steam flow ature (10–15 °C) the AC fan must work at full load.
 G. Barigozzi et al. / Applied Energy 88 (2011) 1366–1376 1375

Fig. 10. DH heat load and Tamb influence on HP turbine exhaust pressure (W&D).

Fig. 13. Optimized normalized net power output: influence of (a) DH heat load and
(b) RH – W&D.
Fig. 11. AC fan load and Tamb influence on normalized net power output – dry.

Fig. 12. Optimized cooling water flow pumps load and CT and AC fans load for
different Tamb – W&D. Fig. 14. Optimized AC fan load for different Tamb – dry.
1376 G. Barigozzi et al. / Applied Energy 88 (2011) 1366–1376

6. Conclusions ing variable steam conditions and their effect on cycle power
output.
A detailed simulation of a wet and dry condensing system in-
stalled in a steam power cycle for district heating was developed. References
A parametric analysis was carried out in order to check the influ-
[1] Conradie AE, Kroger DG. Performance evaluation of dry-cooling systems for
ence of ambient conditions, DH heat demand and condensation
power plant applications. Appl Therm Eng 1996;16:219–32.
heat transfer (i.e. AC, water cooled condenser and CT fans and [2] Smrekar J, Oman J, Sirok B. Improving the efficiency of natural draft cooling
pumps rotational speeds) on plant thermodynamic performances, towers. Energy Convers Manage 2006;47:1086–100.
and particularly, on net power output. No superposition rules [3] El-Wakil MM. Powerplant technology. NewYork: McGraw-Hill; 1984.
[4] Mohiuddin AKM, Kant K. Knowledge base for the systematic design of wet
could be applied, but several parameters simultaneously influences cooling towers. Part I: Selection and tower characteristics. Int J Refrig
in a combined way the net power output. For example, the way the 1996;1:43–51.
CT fan load influences the net power output depends on the ambi- [5] Fisenko SP, Petruchik AI. Toward to the control system of mechanical draft
cooling tower of film type. Int J Heat Mass Transfer 2005;48:31–5.
ent temperature as well as on the AC fan load. Thus, a wide data [6] Brin AA, Petruchik AI, Fisenko SP. Mathematical modeling of evaporative
base of cycle thermodynamic performances was carried out, cover- cooling of water in a mechanical-draft tower. J Eng Phys Thermophys
ing all the possible operating conditions. This data base was in- 2002;75:1332–8.
[7] Fisenko SP, Brin AA, Petruchik AI. Evaporative cooling of water in a mechanical
cluded in an in house developed optimization code providing the draft cooling tower. Int J Heat Mass Transfer 2004;47:165–77.
condensing system control strategy giving the maximum net [8] Fisenko SP, Brin AA. Simulation of a cross-flow cooling tower performance. Int J
power output for each ambient condition and DH requirement. Heat Mass Transfer 2007;50:3216–23.
[9] Rafat AW, Masud B. CFD simulation of wet cooling towers. Appl Therm Eng
The air cooled condenser resulted the best way to reject heat if 2006;26:382–95.
the ambient temperature is lower than 15 °C. At higher ambient [10] Milosavljevic N, Heikkila P. A comprehensive approach to cooling tower
temperature, the condensation should exploit the cooling capacity design. Appl Therm Eng 2001;21:899–915.
[11] Soylemez MS. On the optimum performance of forced draft counter flow
of the tower as much as possible while discarding the remaining
cooling towers. Energy Convers Manage 2004;45:2335–41.
heat in the air cooled condenser. This strategy uses the maximum [12] Qureshi BA, Zubair SM. A complete model of wet cooling towers with fouling
available amount of heat rejection in the cooling tower to reduce in fills. Appl Therm Eng 2006;26:1982–9.
the cost of the air condenser. Another result is about the cooling [13] Bloemkolk JW, Van der Schaaf RJ. Design alternatives for the use of cooling
water in the process industry: minimization of the environmental impact from
water pumps rotational speed: it would be simultaneously in- cooling systems. J Cleaner Prod 1996;1:21–7.
creased with ambient temperature. The AC fan load providing the [14] Harte R, Kratzig WB. Large-scale cooling towers as part of an efficient
best performance in terms of net power came out to be between and cleaner energy generating technology. Thin-Walled Struct 2002;40:
651–664.
40% and 80%. The increase in both heat demand from DH and in [15] Ganan J, Rahman Al-Kassir A, Gonzales JF, Macias A, Diaz MA. Influence of the
ambient RH was found to scale downwards the power production cooling circulation water on the efficiency of a thermonuclear plant. Appl
without having an effect on the optimum condition. Therm Eng 2005;25:485–94.
[16] Lindahl PA, Jameson RW. Plume abatement and water conservation with the
The prediction and the optimization of the thermal cycle per- wet/dry cooling tower. In: Proceedings of the Cooling Tower Institute annual
formance have been achieved by fixing flow rate, temperature meeting; 2003.
and pressure of the steam entering the high pressure turbine.
An interesting hint on how to improve the model lies in consider-